Determining the Best Spatial Statistic Model for Spatial Distribution of Artemisia austriacaSpecies

Mapping the spatial distribution of rangeland species is essential for planning and proper management of rangelands. In this study,the spatial variation of Artemisia austriaca density in the southeast faced slops of Sabalan rangelands was investigated. Overall, 151 sites were selected following systematic-random method. Then, inside each of the determined sites, five 1×1m2plots were selected at 10-meterdistances from each other along the transect. In each plot, the density of A. austriacaspecies was measured and the elevation, slop, direction, rainfall and temperature data were extracted for each site. After normalizing the data, an analysis of spatial statistic was carried out to derive spatial correlation among the collected samples. Finally, variogram and co-variogram were extracted and the best model for fitting was chosen. Based on the best fitted models and using Invers Distance Weighted (IDW), Kriging and Cokriging methods, the map of density was produced for each species. Data analyses and mapping were conducted using SPSS19, GS1 and ArcGIS10 software. Accuracy assessment of the results was carried out using cross validation and three methods including: Root Mean Squared Error (RMSE), Mean Bias Error (MBE) and Mean Absolute Error (MAE). The results showed that, after normalization, the best fitted variogram model was the exponential model (R2=0.95) and data had high spatial correlation (C/(C) =0.85). From among the spatial statistical methods, the best method was the disjunctive Cokiriging (RMSE=0.58). According to the produced maps,Artemisia austriaca are distributed on the northwest (36 to 76%), west and southwest parts of the study area with the highest amount of density. The lowest amount of density was mapped on the northeast, south and southeast (